On 01/14/2014 07:57 AM, ali hagigat wrote:
I am using kernel-3.5.7 with xenomai-2.6.3.
I think the scheduler should schedule task1 and task2(for 3 times
alltogether) and then both tasks should be suspended
You overlooked the fact that both tasks have equal priority. With the
SCHED_FIFO policy, task2 resuming task1 does not mean that task1 will
preempt task2 immediately upon rt_task_resume(), this will happen only
when task2 suspends itself upon rt_task_suspend(), and conversely.
So what happens is that both tasks enter a run-away state by waking up
each other indefinitely, with task2 saturating the kernel log with
printk output. This probably ends when the oldest one (task1) is kicked
out by the watchdog.
but as it has
been indicated in output.txt , there are many "inside task2..."
messages (part of them were copied to the file) and then sudenly task1
is killed!!
Xenomai function seems incorrect(though i may have ignored some facts
as usual, i apologize in advance)
No need to apologize, giving explanations and fixing bugs is part of our
self-inflicted job (the other major part is to actually _implement_
those bugs).
However, you can help too in this Sisyphean process by submitting
detailed, accurate and consistent information, e.g.
Jan 14 09:58:47 localhost kernel: inside task2...
Jan 14 09:58:47 localhost kernel: inside task2...
Jan 14 09:58:47 localhost kernel: Clocksource tsc unstable (delta = 3922090473
ns)
Jan 14 09:58:47 localhost kernel: Switching to clocksource jiffies
According to your report, I would have expected a message like "watchdog
triggered -- killing runaway thread 'KTask1'" appearing in the log. If
so, then you should definitely paste it in your description, it's quite
important.
cat /proc/xenomai/sched
CPU PID CLASS PRI TIMEOUT TIMEBASE STAT NAME
0 0 idle -1 - master R ROOT/0
1 0 idle -1 - master R ROOT/1
2 0 idle -1 - master R ROOT/2
3 0 idle -1 - master R ROOT/3
0 0 rt 50 - master Sf KTask1
KTask1 is suspended and KTask2 is gone in this status report, this does
not match the behavior you described.
Sidenotes:
- you are using the native Xenomai 2.x API in kernel space, which is
obsolete, and was already phased out from the upcoming Xenomai 3.x
specs. If you really need kernel-based support, make it a RTDM driver,
and move the application level code to userland, issuing requests to
this driver.
- please, do __check__ the return codes of __any__ API call you happen
to use in your code, particularly in kernel space context, especially
when something looks wrong to you in the behavior you observe. This
helps ruling out common issues, focusing on the real problem.
--
Philippe.
_______________________________________________
Xenomai mailing list
[email protected]
http://www.xenomai.org/mailman/listinfo/xenomai
